1 214 ACADEMIC EMERGENCY MEDICINE MAR 1998 VOL 5/NO 3 Out-of-hospital Spinal Immobilization: Its Effect on Neurologic Injury Mark Hauswald, MD, Gracie Ong, MBBS, Dan Tandberg, MD, Zaliha Omal; MBBS I ABSTRACT... Objective: To examine the effect of emergency immobilization on neurologic outcome of patients who have blunt traumatic spinal injuries. Methods: A 5-year retrospective chart review was carried out at 2 university hospitals. All patients with acute blunt traumatic spinal or spinal cord injuries transported directly from the injury site to the hospital were entered. None of the 120 patients seen at the University of Malaya had spinal immobilization during transport, whereas all 334 patients seen at the University of New Mexico did. The 2 hospitals were comparable in physician training and clinical resources. Neurologic injuries were assigned to 2 categories, disabling or not disabling, by 2 physicians acting independently and blinded to the hospital of origin. Data were analyzed using multivariate logistic regression, with hospital location, patient age, gender, anatomic level of injury, and injury mechanism serving as explanatory variables. Results: There was less neurologic disability in the unimmobilized n patients (OR 2.03; 95% CI ; p = 0.04). This corresponds to a ~ 2 chance % that immobilization has any beneficial effect. Results were similar when the analysis was limited to patients with cervical injuries (OR 1.52; 95% CI ; p = 0.34). Conclusion: Out-of-hospital immobilization has little or no effect on neurologic outcome in patients with blunt spinal injuries. Key words: injury; trauma; morbidity; spine; immobilization; back board; emergency medical services; spinal cord. Acad. Emerg. Med. 1998; 5: I Immobilization of the spine in blunt trauma is thought to be a crucial intervention almost as essential as management of the airway. Failure to diagnose and appropriately manage spinal injuries is a major concern for emergency physicians. A large number of papers address immobilization and management of spinal injuries in the... Fmm the University of New Mexico. School of Medicine, Albuquerque, NM. Department of Emergency Medicine (MH, DT); and the University of Malaya Faculty of Medicine, Kuala Lumpur;, Department of Anesthesia (GO) and Department of Allied Health Science (ZO). Received: February 19, 1997; revision received: July 10, 1997; accepted: July 17, 1997; updated: October Address for correspondence and reprints: Mark Hauswald, MD, Department of Emergency Medicine, Ambulatory Care Centec 4-%! University of New Mexico, School of Medicine, Albuquerque, NM Fax: ; emergency setting. Much is now known about these issues. Immobilization is improved by using a firm surface; addition of a hard cervical collar? head blocks? and lat- eral provides progressively more stability. The clinical importance of immobilization remains unknown. That is, how much spinal motion is permissible without harm during transport and during the initial workup remains unknown. This issue is complex. The definition of instability is not standardized. The most conservative view is:... the loss of the ability of the spine under physiologic conditions to maintain relationships between vertebra in such a way that there is neither damage nor subsequent irritation to the spinal cord or nerve root and, in addition there is no development of incapacitating deformity or pain from structural changes. 6 This definition, while appropriate to guide long-term management, is of little use in the emergency setting, where the question generally is: will motion
2 Spinal Immobilization, Hauswald et al. 215 make the neurologic lesion worse? Furthermore, neurologic lesions are dynamic, some deteriorate due to swelling and microvascular injury? and some improve as edema and neuropraxia resolve, irrespective of immobilization. Other neurologic injuries are irrevocable at the time of the injury, and not affected by subsequent movement. In the face of these uncertainties and considerable medicolegal pressure, physicians have opted for extraordinarily conservative management. Patients are fully immobilized at the injury site if there is any suggestion that the neck or back could be injured. Immobilization is usually continued in the ED until the spine is cleared by multiple imaging pr~cedures.~* ~ Authors have claimed that without adequate long-term immobilization, 10% to cluded. Patients who died were included unless the cause 25% of all patients with spine injuries will deteri~rate.~ of death was clearly unrelated to the spinal injury; these These claims, however, have little scientific support. cases were almost exclusively patients with massive head Conservative treatment is not necessarily benign. Im- or other injuries who died in the first 24 hours. mobilization is unc~mfortable,l~* ~ takes time, and delays During the period 1988 through 1993, approximately transport. Immobilized patients are difficult to examine 12,700 trauma patients were admitted to inpatient services and treat. Immobilization increases the risk of aspiration at the U.S. hospital and to the University of Maand pressure sores. Cervical collars increase intracranial laysia. Both statistics include transfer patients. The U.S. pressure.14 Given these problems, it would be useful to figures exclude patients with bums, drownings, and isoknow how often not immobilizing patients would result lated injuries who were admitted to services other than in increased neurologic injury. A low incidence of these the trauma service. The n data include the latter acutely unstable injuries would justify more liberal cases. All the patients taken to the U.S. hospital, but none guidelines for allowable spinal motion following trauma. of those taken to the n hospital, had their spines A high incidence of injuries that might benefit from im- immobilized at the injury site. The catchment area of the mobilization would require more a conservative approach. University of Malaya Hospital lacks emergency ambu- Some spinal injuries are undoubtedly truly biome- lance coverage. The hospital operates an ambulance, but chanically and neurologically unstable and will develop it is used almost exclusively for medical patients. Trauma increased neurologic injury with movement. Others are patients are transported by passersby, police, and coworkundoubtedly biomechanically stable but neurologically ers, none of whom have training in spinal immobilization. fragile; these will suffer more neurologic injury by delay- None of the ED staff could remember any patients who ing resuscitation. Standard practice assumes that immo- had been immobilized in. Other differences bebilization is generally protective and that patients with tween our sites are small. The level of training of Malayspinal fractures will have a higher incidence of neurologic sian physicians is comparable to that of their counterparts injuries if immobilization is not carried out. However, this in the, particularly in the essential specialhypothesis has never been tested. It is no longer possible ties where training was commonly outside of to derive a meaningful estimate of effect of spinal im- until recently. The 2 hospitals have similar radiologic, remobilization in the developed world because of the uni- suscitative, and surgical abilities. All patients who were versal adoption of early, preventive immobilization and admitted to either facility after June 1990 with a neurowidespread publicity regarding the need to protect the logic deficit were treated with high-dose methylprednispine until ambulance personnel arrive. We derive this es- so~one. ~ timate by comparing the percentages of spine-injured patients who had neurologic injuries from 2 sites: the University Hospital, University of Malaya in Kuala Lumpur,, which is not served by an out-of-hospital emergency medical services (EMS) system, and the University of New Mexico Hospital in Albuquerque, NM, which is served by an extensive EMS system. I METHODS Study Design: A retrospective chart review of all patients admitted to the inpatient service or ED of our 2 hospitals with spinal or spinal cord injuries between January 1988 and January 1993 was performed. Permission for the study was provided by the Ethical Sub-committee of the Medical Advisory Board of the University Hospital, Kuala Lumpur,, and by the institutional review board of the University of New Mexico, School of Medicine, Albuquerque, NM. Setting and Population: Study cases were identified by searching for bony spine or spinal cord injuries by International Classification of Disease Version 9 (ICD-9) codes contained in hospital computerized databases. Compression fractures due to osteopenia or other disease were ex- Study Protocol: All patients with blunt injuries to the spine or spinal cord who were transported directly from the injury scene to a study hospital were entered into the database. Compression fractures due to osteopenia or disease were excluded. Information regarding hospital, patient age. gender, level of deficit, mechanism of injury, and type of neurologic injury was collected. Ages were grouped by decade for use in the regression model. The level of injury was classified into cervical, thoracic, or lumbosacral depending on the highest vertebra injured. The mechanism of injury was grouped into 1 of 4 cate-
3 216 ACADEMIC EMERGENCY MEDICINE MAR 1998 VOL 5/NO 3 I TABLE 1 Anatomic Distribution of Injuries... Disability No Total Cervical Immobilized (United 34 (30%) 79 (70%) 113 (100%) States) Unimmobilized 10 (25%) 30 (75%) 40 (100%) () Thoracic Immobilized (United 22 (21%) 85 (79%) 107 (100%) States) Unimmobilized 2 (6%) 31 (94%) 33 (100%) () Lumbosacral Immobilized (United 14 (12%) 99 (88%) 113 (100%) States Unimmobilized 1 (2%) 46 (98%) 47 (100%) () gories: falls from a height; motor vehicle crashes (MVCs); high-velocity-low-mass impacts (primarily patients assaulted with blunt objects and those struck by falling objects): and other. The dependent variable, neurologic injury, was classified as disabling or not disabling based on the last hospital note. Patients with complete quadriplegia or paraplegia, inability to ambulate without assistance, incontinence, or the need for chronic catheterization, and those who died were classified as having disability. Patients with no neurologic injury were classified as not having disability. The remaining charts were reviewed by 2 physicians acting independently and blinded to the hospital of origin. These patients were classified into the 2 groups based on whether the physicians thought the injury would interfere with normal functioning. Data Analysis: Comparison between patients from the (all who had spinal immobilization) and (none of whom had spinal immobilization) was performed using x2 and 1-way analysis of variance as appropriate. Multivariate logistic regression of the association between the collected variables and disability was used for analy~is.'~." The level of deficit and the mechanism of injury were coded as separate binary variables. All of the independent variables were included in the model. Odds ratios (ORs) and 2-sided 95% confidence intervals (CIS) were calculated. We also repeated the analysis using only patients with cervical injuries. Data management was carried out using Quattro Pro version 5.00 spreadsheet software (Borland International, Scotts Valley, CA). Statistical computations were performed with Statgraphics Plus version 7.0 (Manugistics Inc., Rockville, MD) and LogXact-Turbo version 1.1 (Cytel Software Corporation, Cambridge, MA). We used 2-tailed tests and an a of 0.05 throughout. I RESULTS... The anatomic distributions of injuries were similar in the 2 sites and to that published in the literature (Table l)." n and US patients were similar in terms of age and level of injury. Patients in were more likely to be male and to have been injured in a fall rather than an MVC (Table 2). There were 24 patients who had injuries that required physician classification. The 2 physicians grouped these with complete agreement (Table 3), resulting in 21% of the patients (70/334) from the and 11 % of the n patients (13/120) being classified as having disabling injuries. The OR for disability was higher for patients in the (all with spinal immobilization) after adjustment for the effect of all other independent variables (2.03; 95% CI ; p = 0.04). The estimated probability of finding data as extreme as this if immobilization has an overall beneficial effect is only 2%. Thus, there is a 98% probability that immobilization is harmful or of no value. The level of neurologic deficit was the only independent predictor of bad outcome (Table 4). We repeated this analysis using only the subset of patients with isolated cervical level deficits. We again failed to show a protective effect of spinal immobilization (OR 1.52; 95% CI ; p = 0.34). I DISCUSSION These results undoubtedly seem counterintuitive to most physicians who have been taught that spinal motion I TABLE 2 Characteristics of the Patients from the and Immobilized Unimmobilized p-value Number of patients Average age 34 yr 35 yr Gender-male 256 (77%)* 106 (88%) Level of injury 0.52 Cervical 113 (34%) 40 (33%) Thoracic 107 (32%) 33 (28%) Lumbosacral 113 (34%) 47 (39%) Mechanism Fall 66 (20%) 63 (53%) Vehicle crash 248 (74%) 45 (38%) Low-mass 9 (3%) 8 (7%) impact Other 11 (3%) 4 (3%) O.OOO1 Significant 70 (21%) 13 (11%) 0.02 disabi I i ty *Percentages are relative to each hospital's total.
4 ~ ~ Spinal Immobilization, Hauswuld et ul. 217 causes neurologic injury. However, technically only the transfer of energy can physically alter material. Acute neurologic injury occurs when excessive energy is deposited in the spinal cord or its vascular structures. This energy is a product of force multiplied by time. Excessive energy is directly related to the failure strength of the material. Over the length of time experienced during an injurious event, the spine is quite strong and massive amounts of energy are required to fracture or otherwise significantly injure it. The cervical spine will fracture when > N (Newton or meter-kg/sec2, 1 N = pounds of force) is applied; the lumbar spine requires >4,200 N to fracture, even in elder individuals.20 Muscles and ligaments reinforce the bone. Even the spinal cord itself is capable of absorbing significant energy without suffering damage.22 Energy deposition during an injury is a complex process. Subjects ejected from vehicles, the most common cause of disability in our sample, undergo repetitive impacts. In most cases the maximal impact is early in the event as the victim contacts the vehicle structure or the ground. It is presumably at these times that most of the injury is inflicted. Subsequently, multiple impacts occur between the subject and the ground. Even in the simple case of a restrained subject and direct linear deceleration while in a sitting position, the initial acceleration is followed by a series of repetitive oscillatory movement^.'^ In these circumstances the energy deposited by moving the patient after the event will be much less than the energy deposited at the scene by secondary impacts. There are good physical and biomechanical reasons why immobilization immediately after the injurious event has little effect. Movement within the spine s normal range of motion requires little energy and is hence unlikely to result in significant energy deposition to the cord. Even the force generated across the spine by hanging a completely unimmobilized 4-kg head off the end of a stretcher is only equal to approximately 40 N, which is orders of magnitude less than that experienced during the original event. As the spine is moved, changes in force vectors occur. The spinal elements (bone, ligament, muscle, and disc) interact to transfer energy to all the component parts.24 This serves to minimize energy deposition to any one component. When force is applied rapidly, the energy is focused due to wave effects, thus enhancing injury.25 However, the definition of instability that is used to guide long-term care of the patient is based on the risk of gradual slippage due to gravity and active motion. It is hardly surprising that this definition has little relevance in the acute setting when the biomechanical factors are completely different. The difference in neurologic disability between immobilized patients in the and unimmobilized patients in was statistically significant. It may I TABLE 3 Physician-classified Patients-Verbatim Discharge Diagnosis Neurologic Finding Injuries judged not disabling Moderate leg weakness, ambulatory Hypoesthetic thumb Paresthesias only Mild hypaesthesia 1/3 right leg Mild hand weakness Decreased right arm sensation Almost normal at discharge Weak deltoids Weak toe Mild diffuse hypaesthesia Paresthesias Mild weakness left leg Sacral 1 root injury Right foot drop Slight right arm weakness Right arm partial brachial palsy Slight left arm weakness Sensory change, no objective findings Injuries judged disabling Right arm paralysis and anesthesia Severe right arm weakness Right hemiparesis Anesthetic left leg Severe hypoesthesia left leg Complete left cervical plexus injury Location I TABLE 4 Logistic Regression Analysis 95% Odds Confidence Ratio Interval p-value Spinal immobilization Gender-male Age (by decade) Level of injury Cervical O.Oo0 1 Thoracic Lumbosacral Mechanism Fall Vehicle crash Low-mass impact Other be that immobilization increases the risk of neurologic injury secondary to tissue hypoxia, perhaps by delaying resuscitation or perhaps the benefit of immobilization is so small that it is unmeasurable given our sample size. Previous studies have estimated that three fourths of cervical fractures are potentially ~nstable ~.~ based on ra-
5 218 ACADEMIC EMERGENCY MEDICINE MAR 1998 VOL 5/NO 3 diographic criteria. The actual percentage of injuries that are likely to be made worse by lack of immobilization during the immediate post-injury period is much smaller. The risk of neurologic deterioration is greatly exaggerated. LIMITATIONS AND FUTURE QUESTIONS... Our study has several shortcomings. Patients who died at the injury site or during transport are excluded. It is possible that some of these died as a result of high cord injuries, attendant loss of diaphragmatic function, and asphyxia. Most of these cord injuries are probably complete at the time of the injury and many of these patients have other fatal injuries, but it is possible that some partial lesions could have been completed during transport in and resulted in death prior to admission. However, there were no survivors in Albuquerque with complete lesions above C, during this period, either. We did not attempt to match patients for the severity of their nonspinal injuries. The University of Malaya does not routinely use injury severity scores, and retrospective calculation of them would have been difficult. The use of mechanism of injury in our regression analysis partly corrects for this omission as does our entry criteria, which required that adequate energy be deposited to injure the spine. It is possible that the injuries from New Mexico were more unstable or more severe. Indeed, our initial plan was to match injuries from our 2 sites and then compare outcomes. This proved impossible. Spinal injuries are idiosyncratic and no 2 are identical. Many injuries were merely described verbally in the radiologic and discharge notes. The severity of injury was poorly predicted by the description; for example, some compression fractures were associated with severe neurologic injuries, while others caused no neurologic injury at all. Fracture classification schemes are not well standardized and systems of classification are based on estimates of long-term instability which may, as noted above, be unrelated to shortterm stability. Even those injuries that were placed in discrete diagnostic categories were not matchable. The number of patients available for comparison is relatively small. We chose to analyze only patients with injuries to the spine presenting to a single pair of mediumsized hospitals over a 5-year period. Inclusion of patients seen prior to 1988 or at other facilities would increase the differences in hospital treatment in our samples and make direct comparison more difficult. Although resources and clinical capabilities are similar in the 2 hospitals, they are not identical. We doubt that hospital care in is significantly superior to that in the, but if this were the case, it would complicate our analysis. An important source of bias in our study is that only patients who proved to have spinal injuries were entered. The vast majority of trauma patients do not have a spinal injury and hence cannot benefit from spinal immobilization. As a result, our study design would tend to exaggerate any potential benefit of current protocols that require the immobilization of almost all trauma patients. It is doubtful that this study can be duplicated in the future because is now developing an EMS system and considerable publicity has recently been given to spinal immobilization in the mass media. Other population-based studies are urgently needed to confirm our data. Current spinal immobilization protocols have been developed without supporting clinical efficiency data. They may be overly conservative. I CONCLUSION Comparison of spine injury patients from 2 study populations, one with out-of-hospital spinal immobilization and the other without, showed a higher rate of neurologic injury in the immobilized group. Acute spinal immobilization may not have significant benefit for the prevention of neurologic deterioration from unstable spinal fractures. The authors acknowledge the assistance of the late Professor N. Subramanian, MBBS (Calc.), Head, Department of Orthopaedic Surgery, University Hospital,, in the planning and data collection for this study. I REFERENCES 1. Alexander RH, Proctor HJ. Advanced Trauma Life Support Course for Physicians. Chicago, IL Committee on Trauma, American College of Surgeons, 1993, pp Rosen PB, McSwain NE Jr. Arata M. Stahl S, Mercer D. Comparison of two new immobilization collars. Ann Emerg Med. 1992; 21: Chiles BW 111. Cooper PR. Acute spinal injury. N Engl J Med. 1992; Graziano AF, Scheidel EA. Cline JR, Baer IJ.A radiographic comparison of prehospital cervical immobilization methods. Ann Emerg Med. 1987; 16: Podolsky S. Baraff LJ. Simon RR, Hoffman JR, Larmon B, Ablon W. Efficacy of cervical spine immobilization methods. J Trauma. 1983; White AA, Panjabi MM. Clinical Biomechanics of the Spine. Philadelphia, PA: J. B. Lippincott, Tator CH, Fehlings MG. Review of the secondary injury theory of acute spinal cord trauma with emphasis on vascular mechanisms. J Neurosurg. 1991; 75: Tator CH. Review of experimental spinal cord injury with emphasis on the local and systemic circulatory effects. Neurochirurgie. 1991; 37: Davis JW, Phreaner DL. Hoyt DB, Mackersie RC. The etiology of missed cervical spine injuries. J Trauma. 1993; 34: Terregino CA, Ross SE, Lipinski MF, Foreman J, Hughes R. Selective indications for thoracic and lumbar radiography in blunt trauma. Ann Emerg Med. 1995; 26: Reid DC, Henderson R. Saboe L, Miller JD. Etiology and clinical course of missed spine fractures. J Trauma. 1987; 27: Cordell WH, Hollingsworth JC. Olinger ML, Stroman SJ, Nelson
6 Spinal Immobilization, Hauswafd et af. 219 DR. Pain and tissue-interface pressures during spine-board immobilization. Ann Emerg Med. 1995; 26: Chan D, Goldberg R, Tascone S, Chan L. The effect of spinal immobilization on healthy volunteers. Ann Emerg Med. 1994; 23: Raphael JH, Chotai R. Effects of the cervical collar on cerebrospinal fluid pressure. Anaesthesia. 1994; 49: Bracken MB. Shepard MJ. Collins WF, et al. A randomized, controlled trial of methylprednisolone or naloxone in the treatment of acute spinal-cord injury. N Engl J Med. 1990; 322: Kleinbaum DG. Analysis of matched data using logistic regression. In: Kleinbaum DG (eds). Logistic Regression: A Self-learning Text. New York. NY Springer-Verlag pp Hirji KF, Mehta CR, Patel NR. Computing distributions for exact logistic regression. J Am Stat Assoc. 1987; 82: Riggins RS, Kraus JF. The risk of neurologic damage with fractures of the vertebrae. J Trauma. 1977; 17: Maiman DJ, Sances A Jr, Myklebust J, et al. Compression injuries of the cervical spine: a biomechanical analysis. Neurosurgery. 1983; 13: Percy 0. Fractures of the vertebral end-plate in the lumbar spinean experimental biomechanical investigation. Acta Orthop Scand. 1957: 25(suppl): 1-10]. 21. Pintar FA, Myklebust JB. Yoganandan N, Mainan DJ, Sances A. Biomechanics of human spinal ligaments. In: Sances A, Thomas DJ, Ewing CL, Larson SJ, Unterharnscheidt F (eds). Mechanisms of Head and Spine Trauma. Goshen, NY Aloray pp Fernandez E, Pallini R, Marchese E, Talamonti G. Experimental studies on spinal cord injuries in the last fifteen years. Neurol Res. 1991; 13: Stapp JP. Human and chimpanzee tolerance to linear decelerative force. In: Sances A, Thomas DJ, Ewing CL, Larson SJ. Unterharnscheidt F (eds). Mechanisms of Head and Spine Trauma. Goshen, NY Aloray. 1986, pp Noyes FR, DeLucas JL. Torvik PJ. Biomechanics of anterior cmciate ligament failure: an analysis of strain-rate sensitivity and mechanisms of failure in primates. J Bone Joint Surg. 1974; 56(A): Fung YC. The application of biomechanics to the understanding and analysis of trauma. In: Nahum AM, Melvin J (eds). The Biomechanics of Trauma. Norwalk, CT: Appelton-Century-Crofts, 1985, pp O Malley KF, Ross SE. The incidence of injury to the cervical spine in patients with craniocerebral injury. J Trauma. 1988; 28: Wright SW. Robinson GG. Wright MB. Cervical spine injuries in blunt trauma patients requiring emergent endotracheal intubation. Am J Emerg Med. 1992: 10:104-9.